The concentration of 23 elements (Al, As, B, Ba, Ca, Cd, Cr, Cu, Fe, Hg, K, Mg, Mn, Mo, Na, Ni, P, Pb, Si, Sr, Ti, V and Zn) has been determined in 15 red wines by ICP-OES (inductively coupled plasma optical emission spectrometry). The similarities and dissimilarities among the element composition of red wines have been elucidated by principal component analysis (PCA). In order to visualize the significance limit of 95% on the matrices of PC loadings and components, the mean value + twice the standard deviation and the mean value-twice standard deviation have also been added to the original matrix of raw data and PCA has been performed on the modified matrix. The dimensionality of the resulting matrices of PC loadings and components has been reduced to one by cluster analysis and to two nonlinear mapping technique (NLMAP). The elements of the original data matrix differed significantly, when the mean value and the mean value + twice the standard deviation and the mean value-twice standard deviation formed a triade on the cluster dendogram or they formed a circle (center being the mean value, radii determined by the mean value + twice the standard deviation and the mean value-twice standard deviation) not overlapping with another circles on the MLMAP. It was established that the element composition of red wines show considerable differences according to the variety of grape. That is this parameter can facilitate the origin and authenticity of the wine. It was further found that the modified PCA method makes possible the visual evaluation of the significant differences among the variables and observations in PCA.